Ca_V2 channel subtype expression in rat sympathetic neurons is selectively regulated by α₂δ subunits

ABSTRACT

Type two voltage gated calcium (Ca_V2) channels are the primary mediators of neurotransmission at neuronal presynapses, but their function at neural soma is also important in regulating excitability.¹ Mechanisms that regulate Ca_V2 channel expression at synapses have been studied extensively, which motivated us to perform similar studies in the soma. Rat sympathetic neurons from the superior cervical ganglion (SCG) natively express Ca_V2.2 and Ca_V2.3.² We noted previously that heterologous expression of Ca_V2.1 but not Ca_V2.2 results in increased calcium current in SCG neurons.³ In the present study, we extended these observations to show that both Ca_V2.1 and Ca_V2.3 expression resulted in increased calcium currents while Ca_V2.2 expression did not. Further, Ca_V2.1 could displace native Ca_V2.2 channels, but Ca_V2.3 expression could not. Heterologous expression of the individual accessory subunits α₂δ-1, α₂δ-2, α₂δ-3, or β4 alone failed to increase current density, suggesting that the calcium current ceiling when Ca_V2.2 was over-expressed was not due to lack of these subunits. Interestingly, introduction of recombinant α2δ subunits produced surprising effects on displacement of native Ca_V2.2 by recombinant channels. Both α₂δ-1 and α₂δ-2 seemed to promote Ca_V2.2 displacement by recombinant channel expression, while α₂δ-3 appeared to protect Ca_V2.2 from displacement. Thus, we observe a selective prioritization of Ca_V channel functional expression in neurons by specific α₂δ subunits. These data highlight a new function for α₂δ subtypes that could shed light on subtype selectivity of Ca_V2 membrane expression.

KEYWORDS:

• calcium channels
• patch-clamp
• sympathetic neurons

Introduction

Type two voltage gated calcium channels (Ca_V2) are found throughout the central and peripheral nervous systems. The most well-described function of these channels is to mediate presynaptic calcium entry and initiate neurotransmission, but in neurons Ca_V2 channels express in the somatodendritic compartment as well, where they can regulate excitability either directly or via coupling to calcium dependent potassium and chloride channels.^1,4 There are three Ca_V2 subtypes, each defined by the sequence of the Ca_Vα1 pore-forming subunit: Ca_V2.1 (Ca_Vα1A), Ca_V2.2 (Ca_Vα1B) and Ca_V2.3 (Ca_Vα1E). The channel is comprised of four domains that form a calcium selective pore, voltage sensing domains, and regions that open and close the channels. Ca_V2.1 and Ca_V2.2 play dominant roles in neurotransmission in the nervous system, as they directly associate with presynaptic proteins SNAP-25 and syntaxin.^5,6 Ca_V2 channels can also couple to the calmodulin kinase (CaMK) nuclear signaling cascade that occurs in the somatodendritic region of the cell, resulting in increased phosphorylation of cAMP response binding protein (CREB), although with lower sensitivity than Ca_V1 channels.^7,8

The dependency on cellular localization for the function of these channels has prompted extensive research to understand how and where the channel subtypes are trafficked in the various neuronal compartments. To be trafficked to the membrane, the Ca_Vα1 subunit associates with auxiliary subunits β and α₂δ.⁹ There are multiple subtypes of these auxiliary subunits, including β1-4 and α₂δ-1-4.⁴ With the exception of α₂δ-4, which is predominately retinal and auditory, all subtypes of auxiliary subunits are found within the central and peripheral nervous system at varying expression levels.^10,11 Functionally, the β subunit associates with the α1 subunit at the endoplasmic reticulum to ensure proper folding of the channel complex, allowing the complex to be trafficked through the secretory pathway.^12,13 The role of α₂δ is more complex. Co-expression of α₂δ potentiates surface expression of α1/β channel complex in most systems.¹⁴ Additionally, α₂δ and β with α1 may also alter channel biophysical properties such as inactivation and channel gating.⁹ In general, any combination of α₂δ, β and α1 subunit will result in enhanced trafficking of the complex to the membrane.⁹

Incoming calcium is highly buffered in a cell.¹⁵ Thus, regulating the amount of channels that are trafficked to the plasma membrane is essential to assure proper function.¹⁶ In exploring how neurotransmission is affected by recombinant channel expression, Mochida et al. demonstrated that recombinant Ca_V2 subtypes reconstituted synaptic transmission similarly to native Ca_V2 channels in rat superior cervical ganglion.¹⁷ What remained to be identified was how the recombinant channels were limited at the presynapse, as overexpression of the channels did not enhance the synaptic transmission. In 2004, Cao and Tsien hypothesized that “slots,” or expression limits due to some unidentified mechanisms, regulated the relative amount of recombinant channels at the presynapse in a subtype dependent manner.¹⁸ Interestingly, Ca_V2.2 mediated current was not reduced by overexpression of recombinant Ca_V2.1.¹⁹ Others have described a direct mechanism to explain these observations. For example, Hoppa et al. argue that limitations in auxiliary Ca_V2 subunits create a bottleneck for channels to be trafficked from the soma to presynapse. However, clearly specific effects of different α2δ subunits were not observed, and effects of distinct α2δ subunits on specific expressed Ca_V channels were not tested.²⁰

In this study, the goal was to examine Ca_V2 channel expression on the somatic membrane of rat superior cervical ganglion (SCG) neurons. By examining somatic rather than presynaptic channels, we hoped to reduce the variables as presynaptic trafficking requires more protein interactions. Further, we seek to determine if there is a general mechanism for subtype selectivity for functional channel expression. We found that recombinant channel expression in SCG neurons from adult rats have expression level limits that are Ca_V2 subtype specific. The general observations were that recombinant Ca_V2.2 has a “ceiling,” or an upper limit of membrane expression similar to the native expression levels. Conversely, overexpression of Ca_V2.1 and Ca_V2.3 resulted in an increase in the number of channels trafficked to the plasma membrane, thus did not appear to be restricted in the same way as Ca_V2.2. Recombinant Ca_V2.1, but not Ca_V2.3 expression, reduced the amount of native Ca_V2.2 channels on the plasma membrane. Interestingly, we found that displacement appeared to depend on α₂δ subunits in a selective way. Expression of α₂δ-1 permitted Ca_V2.3 to displace native Ca_V2.2 channels while α₂δ-3 protected Ca_V2.2 from displacement by other channels, and only α₂δ-2 was associated with a significant increase in current when Ca_V2.2 was overexpressed, either due to relief of the observed ceiling, or by allowing these channels to be expressed via another pathway. These findings show a novel function of α₂δ subtypes, conferring selectivity on recombinant Ca_V2 channel subtypes for preferential expression on the plasma membrane.

Results

Recombinant Ca_V2.1 and Ca_V2.3, but not Ca_V2.2 increase total ICa

Previously, we showed that heterologous expression of Ca_V2.1 in SCG neurons results in an increase in total calcium current density,³ in agreement with previous reports examining hippocampal neurons.¹⁸ Adult rat SCG neurons natively express Ca_V2.2 and Ca_V2.3, with Ca_V2.2 mediating approximately 60–75% of the calcium current.² Heterologous expression of Ca_V2.1 in SCG neurons produced an increase in current without co-expression of α₂δ and β subunits. This was likely due to the fact that SCG neurons are adult, differentiated neurons that natively express these accessory subunits, which can apparently support recombinant expression of Ca_V2.1. Interestingly however, recombinant Ca_V2.2 expression did not result in enhanced current.³ In the present study, we examined in more detail the selectivity of functional expression of Ca_V2 channels in SCG neurons. Overexpression was achieved by intra-nuclear injection of cDNA encoding the α₁ pore-forming subunit of each Ca_V2 channel (with EGFP) into primary cultures of SCG neurons.²¹ Using whole-cell patch-clamp electrophysiology, the relative amount of functional channels on the plasma membrane in each expression condition was determined by application of 80 msec voltage steps ranging from −80 mV to +60 mV in 10 mV increments from a holding potential of −80 mV. The calcium currents were measured 10 msec after the start of each step, at or near steady-state levels, normalized to cell capacitance and plotted against voltage to generate a current voltage (IV) curve.²² Currents did not appreciably inactivate during this time. Fig. 1A shows raw calcium current trace (I_ca) elicited from −70, −10 and +10 mV steps from cells in each overexpression condition. When Ca_V2.2 (rabbit α_1B) was overexpressed the current response did not change significantly compared to uninjected cells (Fig. 1B) suggesting that there is some limit or ceiling for Ca_V2.2 membrane expression. HEK293 cells transfected with the same Ca_V2.2 plasmid (with β4 and α₂δ-1) resulted in measurable calcium current, demonstrating that a functional channel is properly translated and trafficked to the plasma membrane (Fig. 1E). Expression of recombinant non-conductive Ca_V2.2 (“NCCa_V2.2”) reduced the current density below uninjected levels, suggesting that recombinant Ca_V2.2 protein can express in SCG neurons where it appears to displace at least some native channels either at the level of the plasma membrane or during trafficking (Fig. 1C). Because splice isoforms of exon 37 of Ca_V2.2 can affect expression levels,²³ we examined calcium current densities in cells expressing mouse Ca_V2.2 with the two mutually exclusive exons, 37a and b (mouse α_1B). As expected, mouse Ca_V2.2_37a increased current density while Ca_V2.2_37b did not, consistent with our results using the rabbit clone (Fig. 1D). It should be noted that the 37b splice isoform is the dominant variant expressed in SCG neurons,²³ and the increased current observed with 37a is probably due to enhanced association with the adaptor protein AP-1, likely a separate mechanism than that regulating the Ca_V2.2 ceiling in SCG neurons²⁴ because the 37a isoform is relevant to nociceptive neurons but is not expressed in sympathetic neurons. Thus the Ca_V2.2 ceiling observed in our system is pertinent feature unique to SCG neurons Ca_V2.2 with exon 37b, and is consistent with two separate Ca_V2.2 constructs. Finally, expression of GFP alone did not alter the current compared to uninjected cells (not shown).

Figure 1. Effect of heterologous expression of Ca_V2 channel α1 subunits on total calcium current density in rat SCG neurons. (A), Sample calcium current current traces during 80 msec steps to −70 mV, −10 mV, and +10 mV from a holding potential of -80 mV in SCG neurons with the indicated expression. Scale bars in each represent 2 nA. (B), Current-voltage (IV) relations for SCG neurons (average ± SEM) either uninjected (⦁; n = 20), or expressing Ca_V2.1 (▾; n = 14), Ca_V2.2 (□; n = 18), or Ca_V2.3 (▴; n = 14). (C), Current-voltage (IV) relations (average ± SEM) for uninjected (⦁; n = 5) SCG neurons and cells expressing NCCa_V2.2 (□; n = 10). (D), Current-voltage (IV) relations (average ± SEM) for uninjected (⦁; n = 8) SCG neurons and cells expressing exon 37 splice variants of Ca_V2.2, 37a (♦□; n = 9) or 37b (♦□; n = 9). (E), Current-voltage relationship (I-V curves) for control HEK293 cells (untransfected) (○), and those transfected with β4, α2δ-1, and the same Ca_V2.2 construct used in the experiments described in Figure 1B (n = 3; •).

In contrast to the Ca_V2.2 results, Ca_V2.1 (rat α_1A) and Ca_V2.3 (mouse α_1E) overexpression resulted in significantly enhanced current densities, suggesting that more channels were localized to the plasma membrane when these subtypes were recombinantly expressed (Fig. 1B). In the context of current density values from a large number (>1000) of SCG neurons under similar recording conditions, it seems implausible that the increase in current density occurs by chance or by selection bias (Fig. 2). Together these data suggest that Ca_V2.1 and Ca_V2.3 are not limited by the same expression ceiling as Ca_V2.2 in SCG neurons, and that there likely are separate mechanisms limiting these subtypes in their trafficking and stability at the plasma membrane.

Figure 2. Distribution of whole cell capacitance, whole cell calcium currents, and current density in 1108 SCG neurons under the recording conditions employed in this study. (A), relationship between whole cell calcium current and cell capacitance. Plot shows the value from each cell (X), and a linear regression to all of the data yielding a slope of 36 ± 3 pA/pF. (B), Histograms showing the frequency distributions of whole cell capactiance (red), whole cell calcium current (green) and current density (blue) of the cells in (A). These data were fit to a Gaussian distribution of the form illustrated in the inset. Fits are shown as solid lines and yielded the following values: For capacitance: Y_o = 0.50 ± 0.19; A = 6.0 ± 0.5; X_o = 36.3 ± 1.4; width = 21 ± 2.4. For current: Y_o = 0.17 ± 0.2; A = 5.9 ± 0.5; X_o = 0.05 ± 0.04; width = 0.23 ± 0.05. For current density: Y_o = 0.21 ± 0.15; A = 8.5 ± 0.4; X_o = -2.8 ± 0.14 width = 2.7 ± 0.2.

Recombinant Ca_V2.1 displaces native Ca_V2.2_, but recombinant Ca_V2.3 does not

We next investigated if there was competition between recombinant and native channels similar to that reported for some Ca_V2 channels expressed at presynaptic sites.¹⁹ To determine the amount of each Ca_V2 subtype contributing to the total calcium current, one or more selective pharmacological inhibitors were perfused on the cell while a voltage ramp (−120 mV to 80 mV in 160 msec, V_hold = −80 mV) was applied every ten seconds. These data are depicted using dot plots in which open symbols denote the current density before drug application, and closed symbols denote the current in the presence of the blocker. Lines connect peak current densities from individual cells before and after drug application. Average ± SEM values are shown as horizontal lines with vertical error bars. In cells expressing Ca_V2.3 we observed a large decrease (reduced by 29 ± 7 pA/pF; n = 5) of calcium current in the presence of the Ca_V2.3 specific blocker SNX-482 (1 µM),²⁵ while uninjected cells exhibited a much smaller decrease of current density (reduced by 5.4 ± 1 pA/pF n = 6) by SNX-482 (Fig. 3A). Overexpression of a non-conductive Ca_V2.3 (“NCCa_V2.3”), in which all four calcium coordinating glutamate residues in in the channel pore were mutated to alanine, did not result in a significant enhancement or reduction of the SNX-482 sensitive current (Fig. 3A) compared to uninjected cells, with a reduction of 6.4 ± 1 pA/pF (n = 5) upon SNX-482 application (Fig. 3B). Overall, expression of Ca_V2.3 increased the calcium current density, but the result of the NCCa_V2.3 experiment suggested that recombinant Ca_V2.3 channels did not appear to displace native Ca_V2.3, suggesting that under native conditions these channels are not expressed at or very near a ceiling as Ca_V2.2 appears to be.

Figure 3. Pharmacological dissection of enhanced calcium currents when Ca_V2.3 is expressed in SCG neurons. (A), Upper, Summary of control (open, -) and SNX482 inhibited (filled, +) current amplitudes in uninjected (○, •; n = 6), Ca_V2.3- (n = 5), and NCCa_V2.3-expressing (▽, ▾; n = 5) SCG neurons. Averages ± standard errors are also indicated as black lines with error bars for each group. Lower, Bar graph illustrating average (±SEM) current density inhibited by SNX482 (total – remaining current) in each group. * indicates p < 0.05. (B), Upper, Summary of control (open, -) and ω-conotoxin mVIIA- inhibited (filled, +) current amplitudes in uninjected (○, •; n = 3), Ca_V2.3- (n = 7), and NCCa_V2.3-expressing (▽, ▾; n = 5) SCG neurons. Averages ± standard errors are also indicated as black lines with error bars for each group. Lower, Bar graph illustrating average (±SEM) current density inhibited by ω-conotoxin mVIIA (total – remaining current) in each group. ns indicates that no significant differences were detected. (C), Sample current amplitude time course for an uninjected, control SCG neuron during application of ω-conotoxin mVIIA. Points represent peak inward currents from a ramp voltage protocol from -120 to +80 mV, as in (D). (D), Sample current traces illustrating uninhibited current (**) and ω-conotoxin mVIIA inhibited current (*) from the same cell depicted in (C).

Moreover, recombinant Ca_V2.3 did not reduce the amount of Ca_V2.2 current. As shown in Fig. 3B, no significant change was observed in the amount of current sensitive to 500 nM ω-conotoxin mVIIA, a Ca_V2.2 specific blocker, when Ca_V2.3 was expressed.²⁶ ω-conotoxin mVIIA reduced the current in uninjected cells by 14 ± 1 pA/pF (n = 9). A time course of block in an uninjected SCG neuron is shown in Fig. 3C, with sample ramp current traces from that cell shown in Fig. 3D. The currents illustrated are indicated by asterisks (** = the control, unblocked current, * = current inhibited by ω-conotoxin mVIIA). Application of ω-conotoxin mVIIA reduced the current by 12 ± 3 pA/pF (n = 7) in cells expressing Ca_V2.3 (Fig. 3B). When NCCa_V2.3 was overexpressed a reduction of 14 ± 1 pA/pF (n = 9) by ω-conotoxin mVIIA was observed (Fig. 3B), a value indistinguishable from control and Ca_V2.3 expressing cells. Thus, the amount of Ca_V2.2-specific current is not significantly different between functional Ca_V2.3_, NCCa_V2.3 or uninjected conditions. These data indicate that recombinant Ca_V2.3 does not compete with native channels for membrane expression.

To examine the effects of Ca_V2.1 on subtype specific expression, similar experiments were conducted in Ca_V2.1 expressing SCG neurons using the Ca_V2.1 selective blocker ω-agatoxin IVA²⁷ (500 nM) and ω-conotoxin mVIIA (500 nM). As expected, recombinant Ca_V2.1 overexpression resulted in significantly more Ca_V2.1 (ω-agatoxin IVA sensitive) calcium current, 46 ± 8 pA/pF (n = 6) in Ca_V2.1 expressing cells, from nearly undetectable, 2.1 ± 0.3 pA/pF (n = 6) in uninjected cells (Fig. 4B; p = 0.0003 by Student's t-test). Note that in this figure, the open and closed symbol convention used in the other figures was replaced with all open symbols of varying sizes to better illustrate the many overlapping points, especially under control conditions (Fig. 4A). ω-conotoxin mVIIA, a Ca_V2.2 specific blocker, reduced the current by 13 ± 2 pA/pF (n = 19) in uninjected cells, and by 6.9 ± 2 pA/pF (n = 11) in Ca_V2.1 expressing cells. This represents a significant reduction in Ca_V2.2 current when Ca_V2.1 was expressed (p = 0.049 by Student's t test; Fig. 4D). Therefore in contrast to recombinant Ca_V2.3, recombinant Ca_V2.1 overexpression resulted in reduced Ca_V2.2 current compared to uninjected cells. This suggests that recombinant Ca_V2.1 displaces some native Ca_V2.2 channels, presumably by competing with them for the cellular machinery that regulates their trafficking or membrane stability, or both. In this context, we use the term ‘displacement’ to suggest only that when Ca_V2.1 expression is measurably increased, Ca_V2.2 currents decline. The mechanism of this decline is currently unclear and may occur at the level of trafficking of new channels to the membrane, or removal of existing channels more rapidly. Finally, coexpression of recombinant Ca_V2.2 with Ca_V2.1 prevented displacement of Ca_V2.2 current by Ca_V2.1. These data are illustrated in Fig. 4E and F along with paired (same day) control cells. Note that while the mVIIA-sensitive current in the control cells was slightly elevated compared to the previous controls, the difference was not significant. On average, the mVIIA-sensitive current in 6 control cells was 20 ± 5 pA/pF and in 9 cells expressing both Ca_V2.2 and Ca_V2.1, it was 23 ± 4 pA/pF. These values were statistically indistinguishable.

Figure 4. Pharmacological dissection of enhanced calcium currents when Ca_V2.1 is expressed in SCG neurons. (A), Summary of control (-) and ω-Agatoxin IVA inhibited (+) current amplitudes in uninjected (○; n = 10), and Ca_V2.1- expressing (△; n = 7) SCG neurons. Averages ± standard errors are also indicated as black lines with error bars for each group. (B), Bar graph illustrating average (±SEM) current density inhibited by ω-Agatoxin IVA (total – remaining current) in each group. * indicates p < 0.05. (C), Summary of control (open) and ω-conotoxin mVIIA inhibited (filled) current amplitudes in uninjected (○, •; n = 8), and CaV2.1- expressing (△, ▾; n = 9) SCG neurons. Averages ± standard errors are also indicated as black lines with error bars for each group. (D), Bar graph illustrating average (±SEM) current density inhibited by ω-conotoxin mVIIA (total – remaining current) in each group. (E), Summary of control (open) and ω-conotoxin mVIIA inhibited (filled) current amplitudes in control (○; n = 6), and CaV2.1+CaV2.2- expressing (▽; n = 9) SCG neurons. Averages ± standard errors are also indicated as black lines with error bars for each group. F, Bar graph illustrating average (±SEM) current density inhibited by ω-conotoxin mVIIA (total – remaining current) in each group. * indicates p < 0.05.

Since Ca_V2.1 displaced Ca_V2.2 channels but Ca_V2.3 did not, and both increased total current, we asked whether these recombinant channels competed with one another for membrane localization. Ca_V2.1 and Ca_V2.3 cDNA were co-expressed in equal ratios (50 ng/µL of each cDNA) or at 9:1 Ca_V2.1: Ca_V2.3 (90 ng/µL cDNA and 10 ng/µL cDNA, respectively). Co-expression of Ca_V2.1 and Ca_V2.3 in both cases increased total current density above control levels significantly (ANOVA; p < 0.01; Fig. 5A), as expected. Average total peak current density was 24 ± 3 pA/pF (n = 17), 52 ± 3 pA/pF (n = 10), and 53 ± 9 pA/pF (n = 6), in uninjected cells and those expressing 1:1 Ca_V2.1: Ca_V2.3 and 9:1 Ca_V2.1: Ca_V2.3, respectively (Fig. 5A). Average ω-agatoxin IVA-sensitive current density (the difference in current before and after the toxin was applied) was 2.2 ± 0.2 pA/pF (n = 17) in uninjected cells, 25 ± 3 pA/pF (n = 10) in cells injected with Ca_V2.1: Ca_V2.3 at a 1:1 ratio, and 26 ± 12 pA/pF (n = 6), from 9:1 Ca_V2.1: Ca_V2.3 SCG neurons (Fig. 5B). Co-expression of Ca_V2.1 and Ca_V2.3 together resulted in significantly larger ω-agatoxin IVA-sensitive current compared to uninjected cells (ANOVA, p<0.01). This amount of ω-agatoxin IVA specific current was similar to when Ca_V2.1 was expressed alone (Fig. 4A). When NCCa_V2.3 was coexpressed with Ca_V2.1 (50 ng/µL of each cDNA), peak current density was 33 ± 2 pA/pF (n = 7), not significantly less than 1:1 Ca_V2.1: Ca_V2.3 injected cells or cells expressing Ca_V2.1 alone (Figs. 4 and 5A; ANOVA, p<0.01). Thus we cannot conclude that NCCa_V2.3 channels compete with Ca_V2.1 for membrane expression. However, the total current in cells expressing Ca_V2.1 alone, Ca_V2.3 alone, or both channels together was not significantly different, leaving open the possibility that they may be governed by the same expression ceiling, which is clearly separate from that governing Ca_V2.2. Finally, when both Ca_V2.3 and Ca_V2.1 were overexpressed together, native ω-conotoxin mVIIA-sensitive current was 8.7 ± 1 pA/pF (n = 12), significantly smaller than Ca_V2.2 current in uninjected cells, which was 13 ± 5 (n = 8; p = 0.031 by Student's t-test; Fig. 5D). This suggests that recombinant Ca_V2.3 does not prevent recombinant Ca_V2.1 from displacing native Ca_V2.2 channels, and thus Ca_V2.1 selectively competes with Ca_V2.2 for membrane expression. These data suggest that Ca_V2.2 is governed by an expression ceiling, and Ca_V2.3 by a separate ceiling, while Ca_V2.1 expresses more promiscuously, as it appears capable of competing with and displacing both Ca_V2.2 and possibly Ca_V2.3.

Figure 5. Pharmacological dissection of enhanced calcium currents when both Ca_V2.3 and Ca_V2.1 are expressed in SCG neurons. (A), Summary of control (open, -) and ω-Agatoxin IVA inhibited (filled, +) current amplitudes in uninjected (○, •; n = 17) SCG neurons and those expressing Ca_V2.3 and Ca_V2.1- at 1:1 (⋄, ♦□; n = 5) and 1:9 (⋄, ♦□; n = 5) ratio of cDNA injected, and NCCAV2.3 with Ca_V2.1 (⋄, ♦□; n = 6). Averages ± standard errors are also indicated as black lines with error bars for each group. (B), Bar graph illustrating average (±SEM) current density inhibited by ω-Agatoxin IVA (total – remaining current) in each group. * indicates p < 0.05 compared to uninjected controls. (C), Summary of control (open, -) and ω-conotoxin mVIIA inhibited (filled, +) current amplitudes in uninjected (○, •; n = 8) SCG neurons and those expressing Ca_V2.3 and Ca_V2.1 together (⋄, ♦□; n = 13). Averages ± standard errors are also indicated as black lines with error bars for each group. (D), Bar graph illustrating average (±SEM) current density inhibited by ω-conotoxin mVIIA (total – remaining current) in each group. * indicates p < 0.05 compared to uninjected controls.

Changes in auxiliary subunits do not underlie differences in recombinant channel expression

Next, we sought to investigate potential mechanisms that governed differential channel expression ceilings. Since recombinant expression of Ca_V2.2 did not increase the number of functional channels on the plasma membrane (Fig. 1B), we investigated the limiting mechanism for Ca_V2.2. Expressed channels may differentially alter levels of native auxiliary subunits. For example, if native accessory subunit levels are a limiting factor for Ca_V2 expression, it is possible that recombinant Ca_V2.1 and Ca_V2.3 but not Ca_V2.2 may increase these levels, thereby raising the ceiling. RT-PCR analysis indicated all four β subtypes and three α₂δ subtypes (1-3) were expressed at the message level in SCG neurons (Fig. 6A & B). To determine if auxiliary subunit protein expression was selectively altered, immunocytochemistry (ICC) analysis was conducted to probe expression levels of β4 or α₂δ-1 in uninjected, Ca_V2.2, or Ca_V2.3 expressing cells. To verify that small changes in protein expression would be detectable using ICC, we first performed a time-course experiment in HEK293 cells in which β4 was transfected. Cells were fixed and stained for β4 at three time points post-transfection (4-, 6-, 8-hours), as protein levels were expected to rise. Representative images of fluorescent HEK cell staining are shown in Fig. 7A. An approximately 5.5-fold increase in fluorescence was detected between the 4- and 8-hour time points using ICC (Fig. 7A, lower left). These experiments were corroborated with western blot analysis of HEK293 cells collected in triplicate also at 4-, 6-, and 8 hours post-transfection with β4 (Fig. 7A, right). The isolated protein was run on a 10% acrylamide gel and the protein density was calculated with normalization to GapDH expression. We were able to detect a 1.5-fold change in protein density between the 4- and 8-hour time points (Fig. 7A). In these experiments, the 4-hour time point is quite close to background fluorescence, as virtually no fluorescence is visible. These data suggest that relatively small changes in protein expression were detectable using ICC.

Figure 6. mRNAs for calcium channel β1-4 and α₂δ-1-3 are detectable in RNA isolated from rat SCGs. (A), Agarose gel illustrating results of RT-PCR experiments with specific primers for detection of calcium channel β subunits, as indicated. (B), Agarose gel illustrating results of RT-PCR experiments with specific primers for detection of calcium channel α₂δ subunits, as indicated. Lower panels, Expected product sized for each RT-PCR primer set used in the experiments illustrated above.

To determine if overexpression of recombinant Ca_V2 channels differentially altered native expression of auxiliary subunits in SCG neurons, we injected either Ca_V2.2 or Ca_V2.3 with nuclearly-localized dsRED into isolated SCG neurons. The following day the cells were fixed and stained with antibodies (see Methods). Representative images are shown in Fig. 7B. Neurons overexpressing Ca_V2.2 had a measurable increase in β4 fluorescence compared to uninjected cells (Fig. 7B, right), with a ∼2-fold increase (n = 4) compared to uninjected cells. Similarly, overexpression of Ca_V2.3 also resulted in an increase in β4, with an observed ∼2.53—fold increase in fluorescence compared to uninjected cells (n = 3) (Fig. 7B; p<0.05 for each respective condition compared to uninjected cell, one-way ANOVA). Interestingly, a decrease in α2δ-1 fluorescence was observed when Ca_V2.2 was overexpressed compared to uninjected cells (Fig. 7C; p = 0.021, Student's t-test). With Ca_V2.3 overexpression, an apparent decrease was observed in α₂δ-1 fluorescence, but did not reach significance compared to uninjected cells (n = 4; p = 0.062, by Student's t test) (Fig. 7C). It should be noted that the observed decrease in α₂δ-1 may be due at least in part to masking of the antibody epitope as more Ca_V channels are expressed. Unfortunately, sufficiently high quality antibodies for other channel subunits could not be obtained, so we were limited to analysis of α₂δ-1 and β4. Thus when Ca_V2.2 or Ca_V2.3 was overexpressed, levels of auxiliary subunits α₂δ-1 and β4 were altered similarly. It is therefore unlikely that changes in auxiliary subunit availability underlie the observed differences in Ca_V2 channel membrane expression, suggesting the limited ceiling observed for Ca_V2.2 overexpression is regulated by another mechanism.

Figure 7. Changes in accessory subunits upon heterologous expression of Ca_V2 channels. (A), Quantification of subunit expression using ICC compared to western blotting. Upper left, images showing anti-β4 in HEK293 cells transfected with β4 for 4, 6, and 8 hours, as indicated (see methods for details). Upper right, immunoblot showing anti-β4 staining of HEK cell lysates at 4, 6 and 8 hours post transfection. GAPDH loading controls are also shown below. Lower, Quantification of HEK cell fluorescence (left) and western blot density (right) from HEK cells as in Upper. (B) left, Images showing anti-β4 fluorescence in SCG neurons expressing the indicated Ca_V2 subunit. Right, upper, anti-β4 specificity controls showing western blots of HEK cell lysates from cells transfected with the indicated β subunit, or no transfection, as indicated. Lower, quantification of ICC data, as in left panel, for control SCG neurons (“uninjected”) or expressing Ca_V2.2 or Ca_V2.3, as indicated. (C), left, Images showing anti-α₂δ-1 fluorescence in SCG neurons expressing the indicated Ca_V2 subunit. Nuclear ds-Red fluorescence is also shown, as an indicator of successfully injected cells. Right, quantification of ICC data, as in left panel, for control SCG neurons (“uninjected”) or expressing Ca_V2.2 or Ca_V2.3, as indicated. * indicates p < 0.05, vs. control uninjected cells.

To be certain that these changes, particularly the apparently elevated levels of β4 when Ca_V2 channels were expressed, could not explain the differences in recombinant channel expression described above, we heterologously expressed auxiliary subunits alone in SCG neurons and examined whether these would increase calcium current density. In heterologous systems, overexpression of β increases the number of calcium channels trafficked to the plasma membrane,¹³ while α₂δ expression alone does not increase the number of functional channels.²⁸ It is not clear however, whether similar increases would be evident in neurons that natively express calcium channels with auxiliary subunits. Surprisingly,²⁹ expression of neither β4 alone nor α₂δ-1 alone increased current density in SCG neurons (Fig. 8). However, co-expressing β4 with α₂δ-1 did increase current density, although only modestly compared to that seen with Ca_V2.1 or 2.3 expression, suggesting more native channels were localized to the plasma membrane (Fig. 8). These data provide a positive control for expression of each auxiliary subunit, but may also indicate that under the conditions of this study, calcium channel auxiliary subunit availability is not the primary limiting factor for Ca_V2 channel ceilings. Therefore, it appears unlikely that the observed increase in native β4 levels upon Ca_V2 channel expression is responsible for the selective increase in current upon expression of Ca_V2.3 compared with Ca_V2.2. However, the decrease in α₂δ fluorescence in our experiments, as well as research that has supported that α₂δ is involved with membrane stability and even subcellular trafficking,³⁰ prompted us to investigate whether auxiliary subunit α₂δ could affect the selectivity of the calcium channel membrane expression.

Figure 8. Effect of accessory subunit expression on total calcium current density in SCG neurons. Current-voltage relationship (I-V curves) for control SCG neurons (•), and those expressing β4 alone (□), α₂δ-1 alone (♦□), or both subunits together (□). Number of cells in each group is indicated in parentheses in the inset. * indicates p < 0.05, vs. control uninjected cells.

Recombinant expression of α₂δ-1 and α₂δ-2 results in preferential trafficking of recombinant Ca_V2 to plasma membrane

To begin to examine the mechanism of selective channel displacement, individual α₂δ subunits were expressed alone in SCG neurons, or with specific Ca_V2 channel subtypes, and displacement of Ca_V2.2 channels (ω-conotoxin mVIIA sensitive current) was examined. As noted above, expression of α₂δ-1 alone did not detectably alter total current density. Fig. 9A shows a bar graph summary of some of the data described in the figures above, pooled for clarity, including the current sensitive to ω-conotoxin mVIIA (dark gray) and the insensitive current (light gray such that total bar represents the total current) under the indicated Ca_V channel expression conditions. In addition, results from cells expressing α₂δ-1 alone, or with the indicated Ca_V2 channel are shown in Fig. 9B. Expression of α₂δ-1 alone did not change the apparent number of Ca_V2.2 channels expressed or total current compared with uninjected cells (Fig. 9B). SCG neurons expressing α₂δ-1 (10 ng/µl) with Ca_V2.1 (50 ng/µl) in SCG neurons exhibited increased current density above uninjected cells (** Fig. 9B). Ca_V2.2 mediated current remained quite small in neurons expressing both α₂δ-1 and Ca_V2.1 (Fig. 9B). The ω-conotoxin mVIIA-sensitive current in these cells was significantly reduced from 15 ± 3 pA/pF (n = 10) in cells expressing α₂δ-1 alone to just 5 ± 2 pA/pF (n = 9) in Ca_V2.1/α₂δ-1 expressing cells, though the ω-conotoxin mVIIA-sensitive component of the current was not reduced in cells expressing α₂δ-1 alone compared to uninjected cells. Thus, expression of α₂δ-1 with Ca_V2.1 seemingly supports the ability of Ca_V2.1 to displace native Ca_V2.2. Surprisingly, co-expression of α₂δ-1 with Ca_V2.3 also reduced Ca_V2.2 current. Ca_V2.2 currents (mVIIA Ica, dark bars) were 7 ± 1 pA/pF (n = 11) in Ca_V2.3/α₂δ-1 expressing cells (Fig. 9B). Therefore, α₂δ-1 appears to permit Ca_V2.2 displacement by Ca_V2.3 (p = 0.049 by one-way ANOVA), which does not occur in cells made to express Ca_V2.3 alone. Finally, expression of Ca_V2.2 with α₂δ-1 produced total (30 ± 4 pA/pF) and mVIIA sensitive currents (20 ± 3 pA/pF; n = 11) indistinguishable from control or α₂δ-1 expressing cells. Together, these data indicate that recombinant α₂δ-1 expression enhances Ca_V2.2 displacement by other Ca_V2 channels, but does not alter the Ca_V2.2 ceiling.

Figure 9. Summary of the effect of α₂δ subunit expression alone or with Ca_V2 subunits, on total and ω-conotoxin mVIIA sensitive calcium currents in SCG neurons. (A), Pooled data from previous figures illustrating the current densities (total in light gray and ω-conotoxin mVIIA sensitive in dark gray) in uninjected SCG neurons and those expressing each of the Ca_V2 subunits, as indicated. (B), Corresponding data as in (A), but in SCG neurons expressing α₂δ-1 alone (“no Ca_V”) and with each of the Ca_V2 subunits, as indicated. (C), Corresponding data as in (A), but in SCG neurons expressing α₂δ-2 alone (“no Ca_V”) and with each of the Ca_V2 subunits, as indicated. (D), Corresponding data as in (A), but in SCG neurons expressing α₂δ-3 alone (“no Ca_V”) and with each of the Ca_V2 subunits, as indicated. ** = p < 0.05 (ANOVA) for total current density vs. corresponding “No Ca_V” controls; * = p < 0.05 (ANOVA) for mVIIA-sensitive current vs. corresponding “No Ca_V” controls; # = p < 0.05 (ANOVA) for mVIIA-sensitive current vs. uninjected controls, as shown in panel A. far left.

We then asked whether recombinant expression of other α₂δ proteins could alter channel displacement properties in SCG neurons. α₂δ-2 was expressed in SCG neurons alone and with Ca_V2.1 or Ca_V2.3. Similar to results with α₂δ-1, α₂δ-2 expressed alone was not associated with a significant change in total current density (18 ± 3 pA/pF; n = 7; Fig. 9C). There was however a significant reduction in the amount of current sensitive to ω-conotoxin mVIIA compared to uninjected cells, dropping to 5 ± 2 pA/pF (n = 7). In addition, co-expression of Ca_V2.1 with α₂δ-2 resulted in enhanced total current, and the ω-conotoxin mVIIA-sensitive current was just 6 ± 2 pA/pF (n = 9), similar to the reduced current in cells expressing α₂δ-2 alone (Fig. 9C). Expression of Ca_V2.3 and α₂δ-2 together resulted in 10 ± 2 pA/pF (n = 11) ω-conotoxin mVIIA-sensitive current, not significantly smaller than α₂δ-2 alone (Fig. 9C), but note that this value was reduced from control cells. Interestingly, both the total current (45 ± 7 pA/pF) and the mVIIA-sensitive current (29 ± 5 pA/pF; n = 11) were significantly elevated in cells expressing Ca_V2.2 with α₂δ-2 (Fig. 9C), suggesting that it may raise the ceiling for Ca_V2.2 expression, although the fact that the current was not elevated when α₂δ-2 was expressed alone indicates that the role of α₂δ-2 in governing the ceiling is complex. Thus, these data suggest that α₂δ-2, like α₂δ-1, allows recombinant Ca_V2.1 displacement of native Ca_V2.2 channels but permits elevated Ca_V2.2 channel expression.

When expressed alone, α₂δ-3 did not alter total membrane current (Fig. 9D) or the current sensitive to ω-conotoxin mVIIA compared with uninjected cells. The total current density in these cells was 21 ± 4 pA/pF, and the ω-conotoxin mVIIA sensitive current was 13 ± 3 pA/pF (n = 8; Fig. 9D). When α₂δ-3 was co-expressed with Ca_V2.1, the total and ω-conotoxin mVIIA sensitive current was 47 ± 8 and 14 ± 4 pA/pF, respectively (n = 7) (Fig. 9D). With α₂δ-2 and Ca_V2.3 co-expressed, the total current was 60 ± 8 pA/pF, and ω-conotoxin mVIIA reduced the current by 19 ± 4 pA/pF (n = 10) (Fig. 9D). Thus, the ω-conotoxin mVIIA sensitive current when α₂δ-3 is co-expressed alone or with either Ca_V2.1 or Ca_V2.3 is not significantly altered. Next, α₂δ-3 expressing SCG neurons were also injected with recombinant Ca_V2.2 (Fig. 9D). Perhaps surprisingly, neither the ω-conotoxin mVIIA sensitive current (27 ± 7 pA/pF) nor the total current (40 ± 10 pA/pF; n = 10) was significantly altered compared with uninjected cells or cells expressing α₂δ-3 alone. This was perhaps due to the variability, because clearly the average values were quite high. We confirmed the function of α₂δ-3 by expressing this construct with Ca_V2.2 and β4 in HEK293 cells (not shown), which confirmed a significant increase in current density was observed compared to Ca_V2.2/β4 co-expression, consistent with reports that α₂δ-3 stabilizes Ca_V channel expression on the plasma membrane.²⁸ Thus, our data suggest that α₂δ-3, in contrast with α₂δ-1, appears to “protect” native Ca_V2.2 channels from displacement by recombinant channels for plasma membrane expression, but perhaps does not raise the specific Ca_V2.2 expression ceiling. Overall, our data suggest that each α₂δ uniquely alters the relationship between Ca_V2 channel expression priority in SCG neurons. These data highlight a novel, subtype-specific role for α₂δ proteins in differentially prioritizing membrane expression of various Ca_V2 channels, although a great deal more work will be needed to fully understand the distinct trafficking mechanisms that regulate Ca_V channels and the specific roles of the α₂δ subunits in these processes.

Discussion

The goal of this study was to determine if there were discernible patterns of somatic Ca_V2 channel membrane expression in rat SCG neurons. Many expression and trafficking studies regarding Ca_V2 channels have been limited to presynaptic expression^17,19,.²⁰ With many presynaptic-specific interactions it could be possible that those studies overlooked instances of selectivity between Ca_V2 subtypes that occur outside of the presynapse.

We found that in the time frame of these studies, Ca_V2.2 is limited with respect to the number of channels that functionally express at the plasma membrane, unlike Ca_V2.1 and Ca_V2.3 whose expression can be elevated by the introduction of recombinant channel protein. The selective ability of recombinant Ca_V2.1 and Ca_V2.3 (but not Ca_V2.2) to enhance current density in SCG neurons did not appear to result from changes in expression of native Ca_V auxiliary subunits, at least in a simple way, as both Ca_V2.2 and Ca_V2.3 overexpression appeared to alter the expression of α₂δ-1 and β4 similarly (Fig. 7), and overexpression of individual α₂δ or β subunits alone did not increase current density (Figs. 8 & 9). Unfortunately, due to the limitations of available antibodies, we could not test for changes in every subunit, so we cannot definitively rule that out as a mechanism for the selective effects we measured, but based on the data we do have, this seems unlikely. Since auxiliary subunits do not appear to be limiting when recombinant Ca_V2.2 is overexpressed, there is likely another mechanism limiting channels expressed on the plasma membrane. This subtype is the dominant channel in adult SCG neurons, and primarily induces neurotransmission.¹⁷ Therefore it is plausible that there are additional requirements for this channel to be trafficked to the plasma membrane, or the channel is segregated to be trafficked to the presynapse exclusively. In dorsal root ganglia (DRG) sensory neurons it was established that splice isoforms have differential membrane expression patterns.²³ This is not likely related to the mechanism of enhanced current density observed with Ca_V2.1, as the 37a variant of that channel does not result in larger current densities than the 37b variant (data not shown). Therefore, particular protein interactions with those isoforms are clearly essential to determining the number of channels trafficked to the plasma membrane. A limitation in these proteins could perhaps induce the ceilings observed in this study. There have been some indications of proteins that interact with Ca_V2 channels in a subtype dependent manner, though investigation on their influence on selective membrane expression over other Ca_V2 subtypes needs to be pursued in further detail^31,.³²

Further, we found that Ca_V2.1, but not Ca_V2.3, reduces the amount of native Ca_V2.2 expressed on the plasma membrane. This selective ability of recombinant Ca_V2.1 to displace native Ca_V2.2 was studied in more detail. Interestingly, we found that displacement appeared to depend on α₂δ subunits in a subtype specific way (Fig. 9). While expression of α₂δ-1, α₂δ-2 or α₂δ-3 alone did not alter current density, expression of α₂δ-1 and α₂δ-2 seemed to allow Ca_V2.1 to displace native Ca_V2.2 and α₂δ-1 permitted displacement by recombinant Ca_V2.3 when it was co-expressed. By contrast, expression of α₂δ-3 produced the opposite effect, protecting native Ca_V2.2 channels from displacement by other Ca_V2 channels. These findings demonstrate a novel effect of α₂δ subtypes in conferring preference on Ca_V2 channel subtype functional expression. At this time, it is unclear exactly what the ‘displacement’ process entails, so we can only interpret what we actually measure: that when Ca_V2.1 is expressed, for example, there is less mVIIA sensitive current. One possibility is that the rate of turnover of Ca_V2.2 protein is unchanged, but recombinant Ca_V2.1 competes with native Ca_V2.2 during trafficking for insertion into the plasma membrane. Alternatively, expression of recombinant Ca_V2.1 may get to the membrane by a separate mechanism or pathway, then cause loss of native Ca_V2.2 by some other mechanism, possibly by accelerating removal (reducing the half-life of the existing protein). It will be interesting to better understand the details of these processes in the future.

Trafficking studies on Ca_V2 channels have focused heavily on presynaptic localization.⁶ The presynapse is incredibly small and highly organized,³³ and localization of calcium channels near the neurotransmitter machinery is essential for efficient neurotransmission. Further, certain subtypes are preferentially expressed at the presynapse over others^17,.¹⁹ One hypothesis that describes this subtype selectivity is the “slot” hypothesis. In mouse hippocampal neurons, recombinant Ca_V2.2 overexpression did not increase neurotransmission, measured as excitatory post-synaptic potential (EPSC) amplitude, but overexpression of Ca_V2.2 reduced Ca_V2.1-mediated neurotransmission. Further, overexpression of Ca_V2.1 was not able to displace Ca_V2.2-mediated channels at the synapse. Thus, it was hypothesized that there are Ca_V2.2 -specific “slots” and Ca_V2.1-preferring “slots”, with “slots” being undetermined cellular limitations that regulate the amount of Ca_V2 subtype that mediates neurotransmission at the presynapse. It is very interesting that we observe parallels in the present study. At the somatic membrane, we observe a limitation in the amount of Ca_V2.2_-mediated current, and possibly a selectivity, or preference, for recombinant Ca_V2.1 over native Ca_V2.2 channels. While our experiments were conducted in adult rat SCG neurons that do not express native Ca_V2.1, at least in the cell body, the interplay between subtype selectivity for membrane expression is reminiscent of the results in Cao and Tsien's study. Pointedly, the selectivity preferences are opposite for each respective study, suggesting that there may be cell-specific preferences for Ca_V2 subtype expression, or that selective expression (displacement) depends on different factors at the cell body and the synapse. Interestingly, Hoppa et al.²⁰ confirmed that Ca_V2.1 overexpression in hippocampal neurons did not cause displace Ca_V2.2 channels at the synapse, but did lead to increased current density at the cell body, a finding that mirrors what we report in sympathetic neurons here, lending support to the idea that similar mechanisms are in play in both neuronal subtypes, but with strikingly different outcomes in the two cellular compartments.

At the cell body of SCG neurons, it is tempting to speculate that there are at least two separate ‘compartments’ (not necessarily a physical space but perhaps a combination of trafficking circumstances and physical localization on the plasma membrane) for channel expression because 1) N-channels are the predominant natively expressed channel, but heterologous expression of Ca_V2.2 does not increase current density, and 2) heterologous expression of Ca_V2.3, the other natively expressed channel, can increase current density but does not displace native N-channels. We could tentatively call these the “N” compartment and the “R” compartment. It will be interesting in the future to evaluate what the nature of these compartments may be, and how they are shaped by α₂δ subunits. We may speculate that the N compartment may consist of channels on the soma that are destined for trafficking to synaptic sites, while those in the R compartment will reside only on the somatic membrane. This assumption would be accompanied by certain predictions that could be tested. For example, native Ca_V2.2, but not 2.3 channels would likely be associated with presynaptic proteins such as RIM-1. While the complexity of the results presented here indicate that α₂δ subunits are unlikely to be the sole regulators of channel expression in each compartment, the differential effects on channel expression suggest that they play some regulatory role. Perhaps most interestingly was that expression of each α₂δ subunit resulted in a different profile of channel expression with respect to channel expression and ability to displace native Ca_V2.2 channels. Table 1 shows a preliminary assessment of Ca_V2 channel type expression in each putative compartment based on the data obtained to date. Note that because these assessments are based primarily on 1) displacement of native N-currents and 2) enhancement of total current, they are not thorough enough to rule out alternate models that could explain the data. For example, the increase in mVIIA sensitive current when α₂δ-2 and Ca_V2.2 are coexpressed (Fig. 9C) could be explained by a higher ceiling for the N compartment, or by allowing Ca_V2.2 channels to express in the R compartment. In Table 1, we interpret it as the latter since expression of α₂δ-2 alone does not appear to raise the N compartment ceiling. Regardless, the data do provide evidence that SCG neurons appear to produce, traffic, and/or sort channels in a manner that is influenced by α₂δ subunits in a subtype specific way.

Table 1. Categorization of differential effects on Ca_V2 channel expression by α₂δ subunits.

		No α2δ	α2δ-1	α2δ-2	α2δ-3
	CaV2.1	√	√	√
“N” compartment	CaV2.2	√	√	√	√
	CaV2.3		√
	CaV2.1	√	√	√	√
“R” compartment	CaV2.2			√
	CaV2.3	√			√

A mechanism that explains the subtype selectivity in the “slot” hypothesis at presynapses has yet to be described in detail. The auxiliary subunit α₂δ created a “bottleneck” or limitation in Ca_V2 localization to the presynapse,²⁰ so it is interesting that we found a Ca_V2 selectivity preference based on various α₂δs. Though it is not explicitly demonstrated, it is possible that α₂δs are not only the limiting factor in presynaptic membrane localization, but also the mechanism by which selectivity is conferred for presynaptic Ca_V2 subtype expression.

Many circumstances could arise to employ the α₂δ-1 selectivity mechanism for Ca_V2 expression. For example, α₂δ-1 can serve as the receptor for thrombospodin,³⁴ a protein involved in developing synapses, or LRP-1 to influence Ca_V channel trafficking.³⁵ It was also theorized that α₂δ-1 serves as an initial organizer for the synapse.³⁴ It may be possible then the Ca_V2.1 is selectively chosen for presynaptic expression in newly developed or developing synapses. For example, in SCG neurons cultured from P0 rats, Ca_V2.1 is present at the presynapse along with Ca_V2.2. Thus, α₂δ-1 may permit the expression of Ca_V2.1 at the presynapse. However, to date preferential association of α₂δ subunits with Ca_V channels has not been demonstrated. Thus, we propose that the selectivity may arise in a more limited, but still important circumstance. One of the known roles of α₂δ subunits in regulating Ca_V channels is to promote recycling from the endosomal compartment.³⁰ Because α₂δ associates with the extracellular regions of the channel, it would reside within the unique low pH environment inside the endosome. It is possible that in this environment, selectivity of α₂δ-Ca_V channel association is more pronounced and could underlie the phenomena we observe here in which specific α₂δ subunits appear to selectively prioritize Ca_V channel membrane expression. Since only a small fraction of channel complexes are expected to reside in the endosome at any given time, this kind of limited selectivity may have eluded detection using standard biochemical methods. A more thorough test of this hypothesis will be a priority in future studies.

Materials and methods

SCG isolation

The neuronal isolation and injection procedures are from established protocols. SCG were dissected from adult male Wistar rats (175-200 g) and incubated in Earle's balanced salt solution (Life Technologies, Rochelle, MD) containing 0.55 mg/ml trypsin (Worthington, Freehold, NJ), and 1.5 mg/ml Type IV collagenase (Worthington) for 1 hour at 35°C. Cells were collected through centrifugation and transferred to minimum essential medium (Fisher Scientific, Pittsburgh, PA). The cells were then plated, and incubated at 37°C until cDNA injection. All animal protocols were approved by the University of Rochester's Committee on Animal Resources (UCAR).

cDNA injection

Injection of cDNA occurred roughly 4–6 hours after cell isolation. cDNA injections were performed with an Eppendorf 5247 microinjector and Injectman NI2 micromanipulator (Madison, WI). Plasmids were stored at −20°C as a 1–2 μg/μl stock solution in TE buffer (10 mM TRIS, 1 mM EDTA, pH 8). Cells were then incubated at 37°C following injection and experiments are performed the following day.

The rabbit α1B (Ca_V2.2) clone was obtained from Stephen Ikeda (NIAAA, NIH). The exon 37 splice variants for rat α1B were obtained from Addgene (Cambridge, MA) (plasmid #: 26567, 37a; 26569, 37b). The α1A (Ca_V2.1) clone was obtained from Stephen Ikeda (NIAAA, NIH). The human α1E (Ca_V2.3) clone was obtained from Stephen Ikeda (NIAAA, NIH). The rat α₂δ-1 clone was obtained from Stephen Ikeda (NIAAA, NIH). The mouse α₂δ-2 (plasmid #58732) and mouse α₂δ-3 (plasmid #58728) clones were purchased from Addgene (Cambridge, MA). The β4 plasmid was obtained from Stephen Ikeda (NIAAA, NIH). All α1 plasmids were injected at 100 ng/μL unless otherwise indicated. Auxiliary subunits α2δ and β cDNA was injected at 10 ng/μL unless otherwise indicated. All neurons were co-injected with green fluorescent protein cDNA (50 ng/μl; pEGFPC1; Clontech Laboratories, Palo Alto, CA, USA) to identify cells expressing recombinant protein.

Electrophysiology data acquisition and analysis

Patch-clamp recordings were made using 8250 glass (King Precision Glass, Claremont, CA). Pipette resistances were 0.8-3 MΩ yielding uncompensated series resistances of 1–5 MΩ. Series resistance compensation of ≥ 80% was used in all recordings. Data was recorded using an Axopatch 200B patch-clamp amplifier (Molecular Devices, Sunnyvale, CA). Voltage protocol generation and data acquisition were performed using custom procedures written for the Igor Pro software package (Wavemetrics, Lake Oswego, OR) generously provided by Stephen R. Ikeda (NIH, NIAAA) on a Mac OSX Intel Duocore computer with an Instrutech ITC-18 data acquisition board (HEKA Instruments, now Harvard Bioscience, Holliston, MA). Currents were sampled at 100 kHz, low-pass filtered at 5 kHz, and digitized. Data was stored and analyzed offline. All currents were leak subtracted off line prior to data analysis. All patch-clamp experiments were performed at room temperature (21–24°C). The external (bath) calcium current recording solution contained (in mM):145 tetraethylammonium (TEA) 140methanesulfonate, 10 4-(2-Hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 15 glucose, 10 CaCl₂, and 300 nMtetrodotoxin, pH 7.4, osmolality 320 mOsm/kg. The internal (pipette) solution contained: 120 N-methyl-Dglucamine (NMG), 20 TEA, 11 EGTA, 10 HEPES, 10 sucrose, 1 CaCl₂, 4 Mg-ATP, 0.3 Na₂-GTP, and 14 tris-creatine phosphate, pH 7.2, osmolality 300 mOsm/kg.

Data analysis was performed using Igor Pro software (WaveMetrics, Lake Oswego, OR). Using whole-cell patch-clamp electrophysiology, the total amount of functional channels on the plasma membrane in each expression condition was determined by application of 80 msec voltage steps ranging from −80 mV to +60 mV in 10 mV increments from a holding potential (V_hold) of −80 mV. The calcium currents were measured 10 msec after the start of each step, at or near steady-state levels, normalized to cell capacitance and plotted against voltage, generating a current-voltage (IV) curve. Currents did not appreciably inactivate during this time. To determine the amount of each Ca_V2 subtype contributing to the total calcium current, one or more selective pharmacological inhibitors were perfused on the cell while a voltage ramp (−120 mV to 80 mV in 160 msec from a holding potential of −80 mV) was applied every ten seconds until maximum inhibition was achieved. The relative amount of current blocked by that inhibitor represented subtype population functionally expressing on the plasma membrane. Non-specific binding was blocked by applying cytochrome-c from equine heart (90+%) (1 mg/mL; Alfa Aesar, Haverhill, MA). Additionally, all drugs were diluted into external calcium solution with cytochrome-c (1 mg/mL). Concentrations of pharmacological agents perfused were as indicated in the text. Cells were excluded if the leak current at −80 mV was >10% of the peak inward calcium current during a voltage ramp.

Molecular biology and RT-PCR

Ca_V2.2 and Ca_V2.3 were rendered non-conductive through site directed mutagenesis using PfuUltra II Hotstart PCR MasterMix (Agilent Technologies, Santa Clara, CA) followed by DpnI digestion. The mutations done for Ca_V2.2 in the mouse α1B37B exon clone were at amino acids E344A, E693A, E1392A and E1682A. Mutations in human Ca_V2.3 α1E clone were at amino acids E309A, E657A, E1374A and E1665A.

Total RNA from adult rat SCG neurons purified using RNAeasy (Qiagen, Hilden, Germany). RT-PCR was done usingOneStep RT-PCR (Qiagen) following manufacturers protocols. PCR products were run on 2% agarose gel (ThermoFisher Scientific, Waltham, MA) diluted into 1X TAE Electrophoresis Buffer (ThermoFisher Scientific, Waltham, MA). Primers used for β were: β1 forward (5′) AGGGCTCAGCAGAGTCC, β1 reverse (5′) AGAAAGCCAAGACCAAACC, β2 forward (5′) GCATGTTAAGGAAAAATTTAATAATG, β2 reverse (5′) AAGAAGACAGAGCACACTCC, β3 forward (5′) AGGCGGGTTCAGCCGACTC, β3 reverse (5′) ATTTTCTGCACATTAAAGAGAAG, β4 forward (5′) AGCGAAGTCCAAACCTG, β4 reverse (5′) ATGAGGTAACAGACATGATGC. Primers used for α₂δ were: α₂δ-2 forward (5′) CAAGGACAATCGGAACCTGTTT, α₂δ-2 reverse (5′) TGACGGTAGAGCCTTTGTAGA, α₂δ-1 forward (5′) CAAGTCATGGGTGGATAAAATG, α₂δ-1 reverse (5′) TGCAATTTCAACCAGTTGGCG, α₂δ-3 forward (5′) CACTTCAGGGAGCATTTGGA, α₂δ-3 reverse (5′) GTGGAGATCTGGGTGAAAAA.

HEK cell culture

Human embryonic kidney 293 (HEK 293) cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 5% Fetal Bovine Serum (FBS) (Atlas Biologics, Fort Collins, CO) and 2% Penicillin/Streptomycin (MP Biomedicals, LLC, Solon, OH) in a humidified incubator with 5% CO₂. Cells were passed using 0.25% Trypsin (Thermo Fischer Scientific, Waltham, MA) following standard procedures. HEK293 cells were transfected using polyethylenimine (PEI).

Western blotting

Western blots were performed using polyacrylamide gels containing 30% acrylamide (Biorad Hercules, CA), 10% Ammonium persiflage (APS; Biorad), Temed (Biorad), Separating buffer (1.5M Tris base, pH 8.8, 0.4% SDS) (Boston Bioproducts, Ashland, MA), Stacking buffer (0.5M Tris-HCl, pH 6.8, 0.4% SDS) (Boston Bioproducts). 8 or 10% acrylamide gels were run at ∼100 V for twenty minutes, then 40–80 minutes at 120V until protein was adequately through the gel. Protein product was transferred to nitrocellulose membrane (Biorad) in −20°C at 100 V for one hour. The membrane was blocked with 30% milk (Sturm Foods, Manawa, WI) diluted in TBS (19 mMTris base, pH 7.5, 137 mMNaCl, 2.7 mMKCl, 5% Tween-20 (Biorad), before the primary antibody was applied. Primary antibodies were diluted into TBS with 10% milk. Primary antibody was applied overnight at 4°C while rocking. Following four 5 minute washes with TBS, secondary antibody, also diluted in TBS with 10% milk, was applied. After a hour incubation at room temperature, the secondary antibody was washed four times with TBST. Membranes were scanned on a Li-Cor Odyssey (Lincoln, NE) and data was stored and analyzed offline using Igor Pro software (WaveMetrics, Lake Oswego, OR). For densitometry analysis, lines were analyzed based on density. Values for background were subtracted for each lane. Protein bands were normalized to GapDH signal strength. Antibody specificity was confirmed using HEK293 cells expressing each β subunit separately or mock transfected (not shown), and for α₂δ-1 by examining lysates from neural tissue, which express α₂δ subunits, and adipose tissue, which does not (not shown).

Immunocytochemistry

For analyzing auxiliary subunit expression, neurons were isolated, cultured on cover slips were injected with Ca_V2 α1 subtype of interest with nuclearly-localized ds-Red (acquired from Dr. David I. Yule, University of Rochester) for identification. Procedure for immunocytochemistry for isolated cells was guided by published protocols.³⁶ Following overnight culture, cells were fixed with 5 minute application of −20°C chilled acetone (Mallinckrodt Chemicals, St. Louis, MI) followed by 0.01M glycine (Sigma Aldrich, St. Louis, MO) in 1X phosphate buffered saline (PBS; Corning Life Sciences, Tewksbury, MA) for 5 minutes. After a brief wash with 1X PBS cells were permeabilized with triton-X100 (Sigma Aldrich, St. Louis, MO) for 5 minutes and subsequently washed twice with 1X PBS for 1 minute. 1X PBS with 10% bovine serum albumin (BSA; Sigma Aldrich) was used to block non-specific binding for 30 minutes before overnight application (16-18 hours) of primary antibody. The following day cells underwent four 5 minute washes with 1X PBS, followed by 1 hour incubation with the secondary antibody. After four 5 minute washes with PBS, the coverslips were placed on slides with ProLongGold AntifadeMountant (Molecular Probes) and stored at −80° C until use.

Chemical depolarization was induced with potassium chloride solution, as optimized and described for SCG neurons previously.³⁷ SCG neurons were plated onto coverslips and cultured overnight. To induce depolarization, cells were washed with elevated potassium chloride (KCl) solution ranging from 40–90 mM for 15 seconds, then washed with 5 mMKCl solution for 45 seconds. Cells were then fixed and stained as described. The solution, based on Tyrode's solution, was as follows (in mM): 115 NaCl, 40 KCl, 2 MgCl₂, 2 CaCl₂, 10 Hepes, 10 glucose, pH 7.3), with variation in NaCl and KCl concentrations ensuring adequate osmolarity. Pre-treatment with1 μM SNX-482 (Alomone Labs, Jerusalem, Israel) 500 nM ω-conotoxin mVIIA (Tocris Bioscience, Avonmouth, Bristol, UK) or cytochrome-c from equine heart (90+%) (1 μg/μL; Alfa Aesar, Haverhill, MA) diluted in external calcium solution was performed five minutes prior to depolarization.

Antibodies

Anti-α₂δ-1 rabbit polyclonal antibody (1:200) was obtained from Alomone (catalog #ACC-015). Anti-GapDH rabbit polyclonal antibody (1:2500) was obtained from Abcam (catalog #ab9485, Cambridge, MA). Anti-phospho-CREB (SER133) (87G3) rabbit monoclonal antibody (1:133) was obtained from Cell Signaling Technologies (catalong #9198; Danvers, MA). Anti-β4 mouse monoclonal antibody (1:250) was obtained from Abcam (catalog #ab85788). Alexa Fluor 488 goat anti-rabbit IgG (1:10,000) was obtained from Life Technologies (catalog #A11008; Rochelle, MD). Alex Fluor 488 goat anti-mouse (1:10,000) was obtained from Life Technologies (catalog #A11001). Dylight 800 Anti-rabbit IgG (1:10,000) was obtained from ThermoFischer (catalog #SA5-35571; Rockford, IL).

Imaging

Immunofluorescence images were obtained using a TILL Photonics digital imaging system that was ported to a NikonEclipseTi epifluorescent microscope. Fluorescence was measured using a 40x oil immersion objective lens. Confocal images were obtained using EZ-C1 3.60 software (Nikon, Melville, NY). Cells were excited with argon lasers at wavelengths 488 nm and 543 nm. Signal was detected via dichroic mirror on the image plane with 530 ± 30 or 585 ± 30 band pass filter for Alexafluor488 or dsRed, respectively. Images were analyzed with FIJI (Image J). Corrected total cell fluorescence (CTCF) was calculated by measuring the area of the cell, the integrated density of that area and mean fluorescence of the background. The formula used was CTCF = Integrated Density – (Area of selected cell X mean fluorescence of background reading), as described in the literature.³⁸

Statistics

Statistical significance was determined with unpaired Student's t-test (2-tail) for two sample experiments or one-way ANOVA (unweighted) to compare means across multiple samples. P values ≤ 0.05 were considered significant. Group means and standard deviations from experiments were used in Power calculations http://www.biomath.info/power/ttest.htm to utilize statistical tests (a = 0.05 and power = 0.8) estimating the minimum number of cells necessary for each experiment. Curve fitting was done using Igor Pro software (Wavemetrics, Lake Oswego, OR). Linear fits used the equation y = a+bx. Gaussian distributions were fit to y = Aexp[-((x- xo)/width)2].

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Lyndee Knowlton for technical assistance, D. I. Yule (University of Rochester) for use of and assistance with the confocal microscope, R. T. Dirksen, M. E. Dumont, and A. Grossfield (University of Rochester) for helpful discussions.

Funding

Funded by NIH R01 GM101023, to PJK.